Vol. 20, No. 3 Printed in Great Britain
International Journal of Epidemiology ©International Epidemiological Association 1991
Recent Trends in the Incidence of Salivary Gland Cancer PAMELA L HORN-ROSS, DEE W WEST AND SUSAN R BROWN
The causes of salivary gland cancer remain relatively unknown. Only four epidemiological investigations have been reported.1"1 The only established risk factors for salivary gland cancer are radiation exposure,2"9 which accounts for 10% to 30% of disease, and a history of a prior cancer.2JIO~13 Since these two factors appear to have independent effects on the development of salivary gland cancer,23 a history of a prior cancer probably reflects shared risk factors which have not yet been identified for this tumour rather than solely the iatrogenic effects of radiation treatment given for the first cancer.14 Cancer of the salivary gland is rare. Prior to 1985, the age-adjusted incidence rate for this tumour in the US (i.e. as observed by the National Cancer Institute's Surveillance, Epidemiology, and End Results Program) averaged 1.1 per 100 000 among men and 0.8 per 100000 among women. Similar incidence rates were observed in the San Francisco-Oakland Metropolitan Statistical Area (MSA) during this time period. Beginning in 1985, a sudden and sustained doubling of salivary gland cancer incidence, only among men, was observed in the San Francisco-Oakland MSA. The present analysis of cancer registry data was undertaken to determine the nature of this increase. Due to the temporal nature of the increase and its specificity to
men, it was initially hypothesized that the increase was related to the AIDS epidemic. METHODS The data for this analysis were obtained from the San Francisco-Oakland Cancer Registry. This populationbased registry has been part of the Surveillance, Epidemiology, and End Results (SEER) Program since 1973 and part of the California Tumour Registry (CTR) Program since 1987. The registry routinely records demographic and tumour information on all patients diagnosed with cancer who reside in the counties of Alameda, Contra Costa, Mann, San Francisco and San Mateo, California. The cases included in this analysis were defined on the basis of the International Classification of Diseases for Oncology (ICD-O)15 topography codes 142.0142.9. Over 99% of these tumours were microscopically confirmed as primary cancers of the major salivary gland. Incidence rates were calculated using local population projections supplied by the SEER Program. Age-adjusted rates were standardized to the 1970 US population. Linear regression analysis was used to test for trends. Changes in the distribution of patient characteristics between cases diagnosed since the increased incidence was noted, i.e. 1985-1988, and cases diagnosed prior to the increase, i.e. 1973-1984, were assessed using an observed to expected ratio. Estimates of the expected
Northern California Cancer Center, 1420 Harbor Bay Plcwy, Suite 260, Alameda, CA 94501, USA.
628
Downloaded from http://ije.oxfordjournals.org/ at Vanderbilt University on October 13, 2014
Horn-Ross P L (Northern California Cancer Center, 1420 Harbor Bay Pkwy, Suite 260, Alameda, CA 94501, USA), West D W and Brown S R. Recent trends in the incidence of salivary gland cancer. International Journal'ot'Epidemiology1991: 20: 628-633. Beginning in 1985, a sudden and sustained doubling of salivary gland cancer incidence, among men only, is observed in the San Francisco-Oakland Metropolitan Statistical Area. Registry data are examined to determine the nature of this increase and its possible association with the AIDS epidemic. Changes in patient characteristics are assessed by comparing their distribution among recently diagnosed cases (1985-1988) to an expectation based on population growth and the age-specific incidence among patients diagnosed earlier (1973-1984). Based on the observed patterns, it is unlikely that the temporal increase in these tumours is a direct result of the AIDS epidemic or solely the result of a shift in the prevalence of established risk factors. The increase is predominantly seen in men over the age of 75 at diagnosis (O/E = 2.3, p - 0.02) and is observed among both those with and without a prior cancer (O/E - 2.7, p - 0.02 and O/E - 1.5, p = 0.06, respectively). Radiation for the prior cancer was not associated with increased occurrence. Military exposure is crudely approximated by examining birth cohorts. However, the cohort data do not support a hypothesis of military exposure.
INCIDENCE OF SALIVARY GLAND CANCER
number of cases were determined by multiplying the average annual age-specific incidence rates for the period between 1973 and 1984 by the age-specific 1987 populations (the midpoint population for the 19851988 period) for each strata of interest. Chi-square tests were used to assess statistical significance. Calculation of ethnic-specific incidence rates for Hispanic and non-Hispanic Whites in the Bay Area is complicated by the lack of accurate Hispanic-specific intercensil population projections. Hispanic population estimates were therefore approximated by applying the age-specific proportion of Hispanics to non-Hispanics in the 1980 census to the 'total White' population projections for 1987.
1985 it was hypothesized that it might be the related to the AIDS epidemic. The registry does not collect information on comorbid conditions such as AIDS. However, in previous analyses1617 AIDS-related cancers have been identified by examining the demographically defined group of young never-married men residing in San Francisco. Thus, if the increase in salivary gland cancer were AIDS-related, isolated elevations in younger men, San Francisco County residents, and never-married men would be expected. This is not the predominant observed pattern (Table 1). The primary increase is observed in older men (O/E = 2.3, p = 0.02) with increases of borderline statistical significance among men between the ages of 45 and 59 years (O/E = 2.0, p = 0.06). Though the increase in San Francisco County is of borderline statistical significance (O/E = 1.8, p = 0.10), the significant increase is among men over the age of 74 years (O/E = 2.9, p = 0.0004), not younger men (O/E = 1.5, p = 0.31). In Marin County, three times as many salivary gland cases were diagnosed since 1985 as would have been expected. This increase, however, does not reach statistical significance and is based on small numbers of cases. The magnitude of the increase is similar for never-married men and ever-married men (O/E =1.5, p = 0.32 and O/E = 1.6, p = 0.02, respectively). If the increase were AIDS-related, an elevation in the number of Kaposi's sarcomas and non-Hodgkin's lymphomas occurring at this site would also be expected. Only two cases of Kaposi's sarcoma of the
R a t P e r 1 0 0
6 0 0
0
73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88
Year of Diagnosis FIGURE I . Salivary gland cancer age-adjusted incidence rates, San Francisco-Oakland MSA, 1973-1988 (men: , - 0.02; women: pUnMjl •= 0.93).
Downloaded from http://ije.oxfordjournals.org/ at Vanderbilt University on October 13, 2014
RESULTS Figure 1 presents the trends in the annual age-adjusted incidence rates of salivary gland cancer among men and women in the San Francisco-Oakland MSA between 1973 and 1988. The increase in incidence is statistically significant only among men (P = 0.05, F = 7.06, p = 0.02). Prior to 1985, approximately 17 cases of salivary gland cancer in men were reported to the registry annually. Since 1985, this number has increased to 31 cases per year (O/E = 1.6, p = 0.01). For women, the corresponding figures are 16 cases increasing to approximately 20 cases annually (O/ E = 1.3, p = 0.20). The remainder of the analysis is limited to men. Since the increase was confined to men and began in
629
630
INTERNATIONAL JOURNAL OF EPIDEMIOLOGY TABLE 1 Distribution of salivary gland cancer among men, San Francisco-Oakland MSA, 1985-1988
Characteristic Age at diagnosis (years) a; 44 45-59 60-74
Number of cases (1985-1988) Observed Expected
O/E
22.2 15.2 25.7 13.9
08 2.0 1.6 2.3
0.20 3.60 2.58 5.87
0.65 0.06 0.11 0.02
San Francisco San Mateo
34 28 7 32 22
22.9 18.4 2.3 17 9 16.7
1.5 15 3.1 1.8 1.3
1 34 1.25 2.44 2.77 0.42
0.25 0.13 0.12 0.10 0.52
Marital status Never married Ever married
21 101
13.6 63.4
1.5 16
099 5.56
0.32 0.02
32 37 25 9 2 8 10
230 18.3 13.0 6.0 3.2 2.2 11.3
1.4 2.0 1 9 1.5 0.6 3.6 0.9
0.90 479 2.74 0.37 0 11 3.70 0.04
034 0.03 0.10 0.54 0.74 0.05 0.84
Ethnicity Total Whites Non-Hispanic Hispanic Blacks Chinese Japanese Filipino Other
105 97 8 6 5 1 3 2
630 61 2 0.7 5.9 14 0.9 34 0.
1.7 1 6 10.5 1.0 3.5 1.1 0.9 -
698 5.25 17.24 0.00 2.20 0.00 0.01
0.01 002
International Journal of Epidemiology ©International Epidemiological Association 1991
Recent Trends in the Incidence of Salivary Gland Cancer PAMELA L HORN-ROSS, DEE W WEST AND SUSAN R BROWN
The causes of salivary gland cancer remain relatively unknown. Only four epidemiological investigations have been reported.1"1 The only established risk factors for salivary gland cancer are radiation exposure,2"9 which accounts for 10% to 30% of disease, and a history of a prior cancer.2JIO~13 Since these two factors appear to have independent effects on the development of salivary gland cancer,23 a history of a prior cancer probably reflects shared risk factors which have not yet been identified for this tumour rather than solely the iatrogenic effects of radiation treatment given for the first cancer.14 Cancer of the salivary gland is rare. Prior to 1985, the age-adjusted incidence rate for this tumour in the US (i.e. as observed by the National Cancer Institute's Surveillance, Epidemiology, and End Results Program) averaged 1.1 per 100 000 among men and 0.8 per 100000 among women. Similar incidence rates were observed in the San Francisco-Oakland Metropolitan Statistical Area (MSA) during this time period. Beginning in 1985, a sudden and sustained doubling of salivary gland cancer incidence, only among men, was observed in the San Francisco-Oakland MSA. The present analysis of cancer registry data was undertaken to determine the nature of this increase. Due to the temporal nature of the increase and its specificity to
men, it was initially hypothesized that the increase was related to the AIDS epidemic. METHODS The data for this analysis were obtained from the San Francisco-Oakland Cancer Registry. This populationbased registry has been part of the Surveillance, Epidemiology, and End Results (SEER) Program since 1973 and part of the California Tumour Registry (CTR) Program since 1987. The registry routinely records demographic and tumour information on all patients diagnosed with cancer who reside in the counties of Alameda, Contra Costa, Mann, San Francisco and San Mateo, California. The cases included in this analysis were defined on the basis of the International Classification of Diseases for Oncology (ICD-O)15 topography codes 142.0142.9. Over 99% of these tumours were microscopically confirmed as primary cancers of the major salivary gland. Incidence rates were calculated using local population projections supplied by the SEER Program. Age-adjusted rates were standardized to the 1970 US population. Linear regression analysis was used to test for trends. Changes in the distribution of patient characteristics between cases diagnosed since the increased incidence was noted, i.e. 1985-1988, and cases diagnosed prior to the increase, i.e. 1973-1984, were assessed using an observed to expected ratio. Estimates of the expected
Northern California Cancer Center, 1420 Harbor Bay Plcwy, Suite 260, Alameda, CA 94501, USA.
628
Downloaded from http://ije.oxfordjournals.org/ at Vanderbilt University on October 13, 2014
Horn-Ross P L (Northern California Cancer Center, 1420 Harbor Bay Pkwy, Suite 260, Alameda, CA 94501, USA), West D W and Brown S R. Recent trends in the incidence of salivary gland cancer. International Journal'ot'Epidemiology1991: 20: 628-633. Beginning in 1985, a sudden and sustained doubling of salivary gland cancer incidence, among men only, is observed in the San Francisco-Oakland Metropolitan Statistical Area. Registry data are examined to determine the nature of this increase and its possible association with the AIDS epidemic. Changes in patient characteristics are assessed by comparing their distribution among recently diagnosed cases (1985-1988) to an expectation based on population growth and the age-specific incidence among patients diagnosed earlier (1973-1984). Based on the observed patterns, it is unlikely that the temporal increase in these tumours is a direct result of the AIDS epidemic or solely the result of a shift in the prevalence of established risk factors. The increase is predominantly seen in men over the age of 75 at diagnosis (O/E = 2.3, p - 0.02) and is observed among both those with and without a prior cancer (O/E - 2.7, p - 0.02 and O/E - 1.5, p = 0.06, respectively). Radiation for the prior cancer was not associated with increased occurrence. Military exposure is crudely approximated by examining birth cohorts. However, the cohort data do not support a hypothesis of military exposure.
INCIDENCE OF SALIVARY GLAND CANCER
number of cases were determined by multiplying the average annual age-specific incidence rates for the period between 1973 and 1984 by the age-specific 1987 populations (the midpoint population for the 19851988 period) for each strata of interest. Chi-square tests were used to assess statistical significance. Calculation of ethnic-specific incidence rates for Hispanic and non-Hispanic Whites in the Bay Area is complicated by the lack of accurate Hispanic-specific intercensil population projections. Hispanic population estimates were therefore approximated by applying the age-specific proportion of Hispanics to non-Hispanics in the 1980 census to the 'total White' population projections for 1987.
1985 it was hypothesized that it might be the related to the AIDS epidemic. The registry does not collect information on comorbid conditions such as AIDS. However, in previous analyses1617 AIDS-related cancers have been identified by examining the demographically defined group of young never-married men residing in San Francisco. Thus, if the increase in salivary gland cancer were AIDS-related, isolated elevations in younger men, San Francisco County residents, and never-married men would be expected. This is not the predominant observed pattern (Table 1). The primary increase is observed in older men (O/E = 2.3, p = 0.02) with increases of borderline statistical significance among men between the ages of 45 and 59 years (O/E = 2.0, p = 0.06). Though the increase in San Francisco County is of borderline statistical significance (O/E = 1.8, p = 0.10), the significant increase is among men over the age of 74 years (O/E = 2.9, p = 0.0004), not younger men (O/E = 1.5, p = 0.31). In Marin County, three times as many salivary gland cases were diagnosed since 1985 as would have been expected. This increase, however, does not reach statistical significance and is based on small numbers of cases. The magnitude of the increase is similar for never-married men and ever-married men (O/E =1.5, p = 0.32 and O/E = 1.6, p = 0.02, respectively). If the increase were AIDS-related, an elevation in the number of Kaposi's sarcomas and non-Hodgkin's lymphomas occurring at this site would also be expected. Only two cases of Kaposi's sarcoma of the
R a t P e r 1 0 0
6 0 0
0
73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88
Year of Diagnosis FIGURE I . Salivary gland cancer age-adjusted incidence rates, San Francisco-Oakland MSA, 1973-1988 (men: , - 0.02; women: pUnMjl •= 0.93).
Downloaded from http://ije.oxfordjournals.org/ at Vanderbilt University on October 13, 2014
RESULTS Figure 1 presents the trends in the annual age-adjusted incidence rates of salivary gland cancer among men and women in the San Francisco-Oakland MSA between 1973 and 1988. The increase in incidence is statistically significant only among men (P = 0.05, F = 7.06, p = 0.02). Prior to 1985, approximately 17 cases of salivary gland cancer in men were reported to the registry annually. Since 1985, this number has increased to 31 cases per year (O/E = 1.6, p = 0.01). For women, the corresponding figures are 16 cases increasing to approximately 20 cases annually (O/ E = 1.3, p = 0.20). The remainder of the analysis is limited to men. Since the increase was confined to men and began in
629
630
INTERNATIONAL JOURNAL OF EPIDEMIOLOGY TABLE 1 Distribution of salivary gland cancer among men, San Francisco-Oakland MSA, 1985-1988
Characteristic Age at diagnosis (years) a; 44 45-59 60-74
Number of cases (1985-1988) Observed Expected
O/E
22.2 15.2 25.7 13.9
08 2.0 1.6 2.3
0.20 3.60 2.58 5.87
0.65 0.06 0.11 0.02
San Francisco San Mateo
34 28 7 32 22
22.9 18.4 2.3 17 9 16.7
1.5 15 3.1 1.8 1.3
1 34 1.25 2.44 2.77 0.42
0.25 0.13 0.12 0.10 0.52
Marital status Never married Ever married
21 101
13.6 63.4
1.5 16
099 5.56
0.32 0.02
32 37 25 9 2 8 10
230 18.3 13.0 6.0 3.2 2.2 11.3
1.4 2.0 1 9 1.5 0.6 3.6 0.9
0.90 479 2.74 0.37 0 11 3.70 0.04
034 0.03 0.10 0.54 0.74 0.05 0.84
Ethnicity Total Whites Non-Hispanic Hispanic Blacks Chinese Japanese Filipino Other
105 97 8 6 5 1 3 2
630 61 2 0.7 5.9 14 0.9 34 0.
1.7 1 6 10.5 1.0 3.5 1.1 0.9 -
698 5.25 17.24 0.00 2.20 0.00 0.01
0.01 002
